Several years ago, I've installed nvidia graphics drivers using their download files - which require reboots.  I'd prefer to use rpm packages for this and was pleased to find for my system (SL4.1?) and kernel 2.6.9-55.0.2.EL, the legacy driver package (I have a GeForce4 MX 440 with AGP8X),  nvidia-x11-drv-96xx-1.0.9631-1.nodist.rf.i386.  The goal being to enable widescreen resolutions which don't seem to be available under the 'nv' driver.  I'm up to 1400x1050 for a 1680x1050 LCD monitor.  So, I read using nvidia drivers will enable this.

I'm unclear on the sequence of events.  Once I install this package, do I need to reboot?  Is it kernel dependent as is  the case for the original nvidia website drivers?  If so, what kernel is the above yum download for? 

In particular, following the download, when I toggle my xorg.conf settings in /etc/X11 by hand 'nv' to 'nvidia' (re ove 'dri', 'Glcore', add 'glx' in Module section, when I do alt-ctrl-backspace, it goes through a reconstruction of the configuration file as it can't fine modules.  

(EE) Failed to load module "bitmap" (module does not exist, 0)
(EE) Failed to load module "pcidata" (module does not exist, 0)

Apparently, this is do some mishap with the nvidia libglx.so file being placed in /usr/lib/xorg/modules/extensions/nvidia  whereas xorg probably looks in /usr/X11R6/lib/modules/extensions and finds the standard libglx.a file?  I don't want to bother with this.

I'm guessing it is cleaner just to compile the NVIDIA-Linux-x86-96.43.07-pkg1.run file downloaded from nvidia.  Please clue me in if it should be simpler to do this via yum.

Thanks,
Bill Lutter